Daily data population [message #93162] |
Mon, 12 May 2003 07:25 |
viv
Messages: 11 Registered: February 2003
|
Junior Member |
|
|
Hi,
I have a daily job that populates several tables in database A with updated data from a few remote databases(B,C etc.). It's not straight copies but with some logic. The tables are cleared every time and populated with new data. It is done throw procedures and db links. Due to the volume, the job runs quite some time. And I have been experiencing various errors (snap shot too old etc.) recently as data grows. Because of the dependence between tables, fail to populate one table will cause the data intact to be broken.
I was wonder are there alternatives that can handle this better. The most important is to keep data intact and efficiency is a big help too. I am thinking of creating temporary tables in database A to receive the updated data. If all the temporary tables are populated correctly, move the data to the actual tables.
Your suggestions are appreciated and thanks in advance.
|
|
|
|
|
Re: Daily data population [message #93203 is a reply to message #93162] |
Thu, 25 September 2003 13:24 |
Manish
Messages: 79 Registered: December 2000
|
Member |
|
|
First thing i would do try and avoid is doing the transformation across a dblinl.
1) create materialized views on your database( where you load the data) with the same structure as that of the tables in the other database.
2) use fast refresh to populate the materialized views. You can also index the views.
3) to eliminate snapshot too old. use large rollback segments.
please refer Oracle documentation on materialized views for more details.
hth
Manish
|
|
|